from os import listdir
from os.path import isfile, join
import math
import numpy as np
import matplotlib.pyplot as plt
import csv
import copy
import operator
import scipy

# load data


win_rate_matric = {}
win_rate_matric1 = {}

with open('matric.txt') as f:
    for line in f:
        data = line[0:-1].split(' ')
        win_rate_matric[(data[0], data[1])] = eval(data[2])


with open('matric1.txt') as f:
    for line in f:
        data = line[0:-1].split(' ')
        win_rate_matric1[(data[0], data[1])] = eval(data[2])

hero_name_list = []
for key in win_rate_matric:
   if key[0] not in hero_name_list:
      hero_name_list.append(key[0])

# change the ordering of hero_name_list_x
# change the ordering of hero_name_list_y
hero_name_list_x = hero_name_list
hero_name_list_y = copy.deepcopy(hero_name_list)

for t in range(3000):
  # create dict for hero_name_list_x
  dict_x = {}
  for i in range(len(hero_name_list_x)):
    dict_x[i] = 0
    for j in range(len(hero_name_list_y)):
     if hero_name_list_x[i] != hero_name_list_y[j]:
       dict_x[i] = dict_x[i] + j * (win_rate_matric[(hero_name_list_x[i], hero_name_list_y[j])])
  
  sorted_x = sorted(dict_x.items(), key=operator.itemgetter(1))
  temp_hero_name_list_x = []
  for i in sorted_x:
    temp_hero_name_list_x.append(hero_name_list_x[i[0]])
  hero_name_list_x = temp_hero_name_list_x
  # create dict for hero_name_list_y
  dict_y = {}
  for i in range(len(hero_name_list_y)):
    dict_y[i] = 0
    for j in range(len(hero_name_list_x)):
     if hero_name_list_y[i] != hero_name_list_x[j]:
       dict_y[i] = dict_y[i] + j * (win_rate_matric[(hero_name_list_x[j], hero_name_list_y[i])])
  
  sorted_y = sorted(dict_y.items(), key=operator.itemgetter(1))
  temp_hero_name_list_y = []
  for i in sorted_y:
    temp_hero_name_list_y.append(hero_name_list_y[i[0]])
  hero_name_list_y = temp_hero_name_list_y


import sklearn
from sklearn.cluster import KMeans
from sklearn import metrics
from sklearn.cluster import KMeans
from sklearn.datasets import load_digits
from sklearn.decomposition import PCA
from sklearn.preprocessing import scale

x = []
y = []
c = []
s = []
m = []
for i in range(len(hero_name_list_x)):
   temp_m = []
   for j in range(len(hero_name_list_y)):
    if hero_name_list_x[i] != hero_name_list_x[j]:
      temp_m.append(win_rate_matric[(hero_name_list_x[i], hero_name_list_x[j])])
      x.append(len(hero_name_list_x) - i)
      y.append(j)
      if win_rate_matric[(hero_name_list_x[i], hero_name_list_x[j])] > 0:
         rgb = (1.0, 0.0, 0.0)
      else:
         rgb = (0, 0.5, 0.0)
      t = abs(win_rate_matric[(hero_name_list_x[i], hero_name_list_x[j])])/3
      if t > 1:
         t = 1
      c.append((rgb[0], rgb[1], rgb[2], t))
      s.append(5)
    else:
      temp_m.append(0)
   m.append(temp_m)


num_clusters = 10

estimator = KMeans(init='k-means++', n_clusters=num_clusters, n_init=10)

estimator.fit(np.array(m))

cutoff = []
hero_name_list_x_temp = []
estimator.labels_
for i in range(num_clusters):
   for j in range(len(hero_name_list_x)):
      if estimator.labels_[j] == i:
         hero_name_list_x_temp.append(hero_name_list_x[j])
   cutoff.append(len(hero_name_list_x_temp))

hero_name_list_x = hero_name_list_x_temp

cutoff = [0] + cutoff
name_list = []
for i in range(len(cutoff) - 1):
   temp_list = hero_name_list_x[cutoff[i]: cutoff[i+1]]
   name_list.append(temp_list)

result = range(10)
avg_matrix = []
for i in result:
    temp = []
    for j in result:
       sum = 0
       count = 0
       for e in name_list[i]:
          for e1 in name_list[j]:
             if e != e1:
                sum = sum + win_rate_matric[(e, e1)]
                count = count + 1
       temp.append(sum/(count))
    avg_matrix.append(temp)


def search(l, m):
   if len(l) == 10:
      temp = 0
      for i in range(10):
         for j in range(10):
            d_list = [(-1, 0), (1, 0), (0, -1), (0, 1)]
            for d in d_list:
               if d[0] + l[i]< 10 and d[0]+l[i] >= 0 \
               and d[1] + l[j]<10 and d[1]+l[j] >= 0:
                  temp = temp + abs(m[l[i]][l[j]] - m[l[i] + d[0]][l[j]+d[1]]) * (1.0/(abs(i-j)+1)**2)
      return temp, l
   max = 0
   result = []
   for i in range(10):
      if i not in l:
         t, r = search(l+[i], m)
         if t > max:
            result = r
            max = t
   return max, result



temp, result = search([], avg_matrix)

result = [3, 6, 7, 5, 8, 2, 4, 9, 1, 0]

temp = []
for i in range(10):
   temp = temp + name_list[result[i]]

hero_name_list_x = temp

x = []
y = []
c = []
s = []
m = []
for i in range(len(hero_name_list_x)):
   temp_m = []
   for j in range(len(hero_name_list_y)):
    if hero_name_list_x[i] != hero_name_list_x[j]:
      temp_m.append(win_rate_matric[(hero_name_list_x[i], hero_name_list_x[j])])
      x.append(len(hero_name_list_x) - i)
      y.append(j)
      if win_rate_matric[(hero_name_list_x[i], hero_name_list_x[j])] > 0:
         rgb = (1.0, 0.0, 0.0)
      else:
         rgb = (0, 0.5, 0.0)
      t = abs(win_rate_matric[(hero_name_list_x[i], hero_name_list_x[j])])/3
      if t > 1:
         t = 1
      c.append((rgb[0], rgb[1], rgb[2], t))
      s.append(5)
    else:
      temp_m.append(0)
   m.append(temp_m)

plt.scatter(x, y, color=c, s=s)
#for t1 in cutoff:
#   plt.plot(range(-1, 120), [len(hero_name_list_x) - t1]*121, '-')
#   plt.plot([t1]*121, range(-1, 120), '-')

plt.show()


for i in result:
 for t in range(3000):
  hero_name_list_y = hero_name_list_x
  # create dict for hero_name_list_x
  dict_x = {}
  for key in name_list[i]:
    dict_x[key] = 0
    for j in range(len(hero_name_list_y)):
     if key != hero_name_list_y[j]:
       dict_x[key] = dict_x[key] + j * (win_rate_matric[(key, hero_name_list_y[j])])
  
  sorted_x = sorted(dict_x.items(), key=operator.itemgetter(1))
  temp_hero_name_list_x = []
  for item in sorted_x:
    temp_hero_name_list_x.append(item[0])
  name_list[i] = temp_hero_name_list_x
  hero_name_list_x = []
  for item in result:
     hero_name_list_x = hero_name_list_x + name_list[item]